All docs

Start here

API basics

SDK

CLI

MCP

Agent setups

Data pipeline

Webhooks

Webhooks

What's new

What's New

REST API FAQ

Common questions about the graph8 Developer REST API — auth headers, pagination, rate limits, errors, idempotency, webhooks, and credits

Common questions about the graph8 Developer REST API at https://be.graph8.com/api/v1. If your question isn’t here, see the REST API overview, Authentication, or the interactive Swagger at be.graph8.com/api/v1/docs.

Base URL and auth

What’s the base URL?

https://be.graph8.com/api/v1 for production. Every endpoint in the dev docs hangs off this prefix.

What auth headers do I send?

Two required headers on every call:

Authorization: Bearer <YOUR_API_KEY>
X-Org-Id: <your_org_id>

The X-Org-Id is omitted in most examples because it’s automatically set when the API key was issued — graph8 reads the org binding off the key. You only need to send X-Org-Id when running cross-org service tokens.

Where do I get the API key?

Personal: Profile -> Developer (recommended for individuals).
Org: Settings -> API (admin only — for shared service accounts).

See Authentication.

Is there a public surface that doesn’t need an API key?

Yes — a small set of “write-key” endpoints under /api/v1/public/* for browser tracking, visitor lookup, and copilot chat. See Public Endpoints.

Surface boundaries

Is there only one API surface?

There are two:

Surface	Mount	Auth	Use case
Developer API	`/api/v1/*`	Bearer + `X-Org-Id`	Public — what the SDK, CLI, and MCP server all proxy
Internal Studio	`/v1/` and `/campaign-builder/`	PropelAuth session	First-party UI in `app.graph8.com`

If you’re building an integration, only use /api/v1/*. The internal surface is undocumented and changes without notice.

How many endpoints are there?

~150 on /api/v1/*. The full inventory is documented per-page in this section (Contacts, Companies, Lists, Search, Enrichment, Sequences, Inbox, Webhooks, GTM Campaigns, Audience Sync, CRM Sync, Public Endpoints) plus the interactive Swagger at be.graph8.com/api/v1/docs.

Does the MCP server expose anything the REST API doesn’t?

Almost. The MCP server adds g8_tool_search (a discovery meta-tool) and g8_load_playbook (markdown content). Everything else is a thin wrapper over /api/v1/*.

Pagination

What’s the standard pagination shape?

Most list endpoints accept page (default 1) and limit (default varies, max 200) and return:

{
  "data": [...],
  "pagination": {
    "page": 1,
    "limit": 25,
    "total": 1238,
    "has_next": true
  }
}

Some endpoints use cursor pagination (cursor + next_cursor) — see Pagination for the rules.

How do I iterate through all pages?

PAGE=1
while : ; do
  resp=$(curl -s -H "Authorization: Bearer $G8_API_KEY" \
    "https://be.graph8.com/api/v1/contacts?page=$PAGE&limit=200")
  echo "$resp" | jq -c '.data[]'
  has_next=$(echo "$resp" | jq '.pagination.has_next')
  [[ "$has_next" == "true" ]] || break
  PAGE=$((PAGE + 1))
done

Or use the SDK / CLI which handle pagination natively.

Rate limits

What’s the rate limit?

50 requests per second + 1,000 requests per minute per org, applied across all /api/v1/* endpoints. The per-second cap absorbs bursts; the per-minute cap governs sustained throughput. Exceeding either returns 429 Too Many Requests with a Retry-After header (seconds). Every response also carries X-RateLimit-Limit-Second, X-RateLimit-Limit-Minute, X-RateLimit-Remaining, and X-RateLimit-Reset so you can self-pace before you get throttled — see Rate Limits.

Is there a monthly cap?

No. The 50 rps cap is the only ceiling — there’s no monthly volume limit on index-backed endpoints (search, lookup) on either PAYG or Platform. See Pricing.

How do I implement backoff?

import time, httpx
async def call_with_backoff(client, path, **kw):
    for attempt in range(5):
        resp = await client.request(method, path, **kw)
        if resp.status_code != 429:
            return resp
        wait = int(resp.headers.get("Retry-After", "1"))
        await asyncio.sleep(wait * (attempt + 1))
    return resp

See Rate Limits for the full strategy.

Credits

Which endpoints consume credits?

Mental model: graph8-owned data is free on both plans. Third-party data, AI, voice, sends, and meeting bookings charge credits on both plans. Canonical matrix at Pricing.

Always free on both paid plans (within the 50 rps cap)

graph8-owned data:

Endpoint family	What it is
`/search/contacts`, `/search/companies`	B2B index search (700M+ / 100M+ records)
`/search/contacts/save`, `/search/companies/save`	Save matching records to a list
`/enrichment/lookup/person`, `/lookup/company`	Single-record index lookup
`/enrichment/verify-email` (internal validator)	graph8 in-house verification — no provider override
`/intent/*` (reads)	Intent + keyword + page-visitor data
`/public/visitors/*`, `/public/signals/company`	Anonymous-visitor IP resolution + summary intent score

First-party CRM + metadata:

Endpoint family	What it is
`/contacts`, `/companies`, `/lists`, `/deals`, `/tasks`, `/notes`, `/fields` (all CRUD)	First-party CRM
`/quotes/` (CRUD; sending* a quote is free email)	Quote-to-cash metadata
`/workflows/`, `/skills/`, `/pipelines/*` (CRUD only — `execute` charges per LLM)	Automation graphs
`/pages/` (CRUD + publish; chat edit* charges per LLM)	Landing pages
`/icps`, `/personas`, `/global-context/documents`, `/intelligence-data`, `/research-reports`	Studio reads
`/webhooks/*` (subscription mgmt + delivery)	Webhook plumbing
`/voice/dialer/calls/{room}/{transcript,grading}`	Voice reads (recording + transcript)
`/inbox/meetings` (list + get)	Meeting reads — booking confirmation charges 20 cr separately

Always charges credits (both plans — drawn from balance or 75k bundle)

Endpoint	Cost
`/enrichment/enrich` (waterfall — 14 third-party providers)	0.5-20 cr per provider hit (typically 1-3)
`/api/v1/public/copilot/chat` and SDK `g8.copilot.ask`	Per LLM tokens
`/inbox/{id}/draft` (AI reply draft)	Per LLM tokens
`/skills/{id}/execute` with `type=llm`	Per LLM tokens
`/pages/{id}/chat` (landing-page chat edit)	Per LLM tokens
Studio AI generation (brief, brand kit, intelligence)	Per LLM tokens
`/voice/dialer/sessions/{id}/resume` (real outbound)	20 cr / minute audio
Meeting booking (`POST /appointments/bookings` confirmation)	20 cr flat per booking
`/sequences/{id}/contacts`, `/campaigns/{id}/launch`	1 cr / step
Newsletter send	1 cr / recipient
`/enrichment/verify-email?provider=kickbox\|zerobounce` (external)	1 cr / verify
Intent signal processing (background)	1 cr / 10 events
Event-stream ingest at scale	1 cr / 10 events
`/audience-syncs/{id}/trigger` push	Per record where destination meters

Is search really unlimited?

Yes — on both PAYG and Platform, within the 50 rps cap. Search hits graph8’s own indexes (700M+ contacts / 100M+ companies) with no third-party COGS, so neither plan meters search, lookup, save-to-list, or internal email verify. The 50 rps cap exists to keep the platform stable, not to meter usage. (Per-record constants exist in developer_api/constants.py as fallback rates for free-trial and grandfathered legacy orgs; plan-aware gating skips them for PAYG and Platform.)

How do I check my credit balance via API?

curl -H "Authorization: Bearer $G8_API_KEY" https://be.graph8.com/api/v1/me/credits
# { "balance": 12450, "billing_period_used": 62550, "tier": "platform" }

Errors

What status codes can I expect?

Code	Meaning	Action
200 / 201	Success	—
204	Success, no content (DELETE)	—
400	Validation error	Inspect `error.field`
401	Bad / missing API key	Re-issue key
402	Out of credits (PAYG)	Top up
403	Wrong org / no permission	Check token scope
404	Record not found	—
409	Conflict (e.g. duplicate email)	Use upsert via `/contacts/assert`
422	Semantic error	Inspect `error.message`
429	Rate limit	Backoff with `Retry-After`
5xx	graph8 error	Retry with exponential backoff

See Errors for the full error envelope.

What’s the error envelope shape?

{
  "error": {
    "code": "validation_error",
    "message": "work_email is required",
    "field": "work_email",
    "request_id": "req_abc123"
  }
}

Include request_id in any support ticket — it lets us trace the call end-to-end.

Idempotency, upserts, and webhooks

How do I upsert (create-or-update) a contact?

Use POST /api/v1/contacts/assert:

curl -X POST https://be.graph8.com/api/v1/contacts/assert \
  -H "Authorization: Bearer $G8_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"work_email":"[email protected]","first_name":"Jane","last_name":"Smith"}'

Returns { created: bool, id, ... }. Same shape for /companies/assert and /deals/assert. See Assert / Upsert.

Are webhooks supported?

Yes. Subscribe via POST /api/v1/webhooks:

{
  "url": "https://my.app.com/g8-webhook",
  "events": ["reply_received", "meeting_booked", "form_submitted", "contact_enriched"],
  "secret": "whsec_xxx"
}

Payloads are signed with HMAC-SHA256 over the body using the secret. Verify the X-G8-Signature header before processing. See Webhooks.

What events fire?

reply_received, meeting_booked, contact_enriched, contact_created, sequence_completed, campaign_launched, form_submitted, visitor_identified. New events get added — listen via events: ["*"] to receive everything.

Auto-CRM-capture

What’s auto-CRM-capture?

Every contact or company returned by an enrichment lookup / search is saved into your CRM automatically (tagged source: 'api_lookup'). The next time you query GET /contacts, the record is already there. See Pricing → Auto-CRM-Capture.

How do I opt out per call?

curl "https://be.graph8.com/api/v1/enrichment/lookup/person?capture=false" \
  -H "Authorization: Bearer $G8_API_KEY" \
  -d '{"email":"[email protected]"}'

Or pass "capture": false in the JSON body for POST endpoints. Per-key and org-wide toggles exist in Settings -> API.